CN108377658B - Autofocus system and method in digital holography - Google Patents
Autofocus system and method in digital holography Download PDFInfo
- Publication number
- CN108377658B CN108377658B CN201580065121.7A CN201580065121A CN108377658B CN 108377658 B CN108377658 B CN 108377658B CN 201580065121 A CN201580065121 A CN 201580065121A CN 108377658 B CN108377658 B CN 108377658B
- Authority
- CN
- China
- Prior art keywords
- depth
- edge
- image
- edge detection
- focal plane
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 166
- 238000001093 holography Methods 0.000 title description 5
- 238000003708 edge detection Methods 0.000 claims abstract description 104
- 238000005259 measurement Methods 0.000 claims abstract description 89
- 239000006185 dispersion Substances 0.000 claims abstract description 23
- 238000012360 testing method Methods 0.000 claims description 12
- 239000000571 coke Substances 0.000 claims description 7
- 230000005855 radiation Effects 0.000 claims description 3
- 238000004422 calculation algorithm Methods 0.000 description 21
- 238000001514 detection method Methods 0.000 description 20
- 238000003384 imaging method Methods 0.000 description 19
- 230000003247 decreasing effect Effects 0.000 description 13
- 230000006870 function Effects 0.000 description 11
- 238000005457 optimization Methods 0.000 description 11
- 238000012545 processing Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 10
- 230000008569 process Effects 0.000 description 10
- 230000008901 benefit Effects 0.000 description 8
- 230000003287 optical effect Effects 0.000 description 7
- 239000007787 solid Substances 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000001914 filtration Methods 0.000 description 5
- 239000011159 matrix material Substances 0.000 description 5
- 230000011218 segmentation Effects 0.000 description 5
- 238000003860 storage Methods 0.000 description 4
- 238000011179 visual inspection Methods 0.000 description 4
- 238000004590 computer program Methods 0.000 description 3
- 230000007423 decrease Effects 0.000 description 3
- 230000001419 dependent effect Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000000386 microscopy Methods 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 230000003252 repetitive effect Effects 0.000 description 2
- 241000406668 Loxodonta cyclotis Species 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 230000001427 coherent effect Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 1
- 239000010931 gold Substances 0.000 description 1
- 229910052737 gold Inorganic materials 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000000813 microbial effect Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 244000045947 parasite Species 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G03—PHOTOGRAPHY; CINEMATOGRAPHY; ANALOGOUS TECHNIQUES USING WAVES OTHER THAN OPTICAL WAVES; ELECTROGRAPHY; HOLOGRAPHY
- G03H—HOLOGRAPHIC PROCESSES OR APPARATUS
- G03H1/00—Holographic processes or apparatus using light, infrared or ultraviolet waves for obtaining holograms or for obtaining an image from them; Details peculiar thereto
- G03H1/04—Processes or apparatus for producing holograms
- G03H1/08—Synthesising holograms, i.e. holograms synthesized from objects or objects from holograms
- G03H1/0866—Digital holographic imaging, i.e. synthesizing holobjects from holograms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G03—PHOTOGRAPHY; CINEMATOGRAPHY; ANALOGOUS TECHNIQUES USING WAVES OTHER THAN OPTICAL WAVES; ELECTROGRAPHY; HOLOGRAPHY
- G03H—HOLOGRAPHIC PROCESSES OR APPARATUS
- G03H1/00—Holographic processes or apparatus using light, infrared or ultraviolet waves for obtaining holograms or for obtaining an image from them; Details peculiar thereto
- G03H1/04—Processes or apparatus for producing holograms
- G03H1/0443—Digital holography, i.e. recording holograms with digital recording means
-
- G—PHYSICS
- G03—PHOTOGRAPHY; CINEMATOGRAPHY; ANALOGOUS TECHNIQUES USING WAVES OTHER THAN OPTICAL WAVES; ELECTROGRAPHY; HOLOGRAPHY
- G03H—HOLOGRAPHIC PROCESSES OR APPARATUS
- G03H1/00—Holographic processes or apparatus using light, infrared or ultraviolet waves for obtaining holograms or for obtaining an image from them; Details peculiar thereto
- G03H1/04—Processes or apparatus for producing holograms
- G03H1/0443—Digital holography, i.e. recording holograms with digital recording means
- G03H2001/0447—In-line recording arrangement
-
- G—PHYSICS
- G03—PHOTOGRAPHY; CINEMATOGRAPHY; ANALOGOUS TECHNIQUES USING WAVES OTHER THAN OPTICAL WAVES; ELECTROGRAPHY; HOLOGRAPHY
- G03H—HOLOGRAPHIC PROCESSES OR APPARATUS
- G03H1/00—Holographic processes or apparatus using light, infrared or ultraviolet waves for obtaining holograms or for obtaining an image from them; Details peculiar thereto
- G03H1/04—Processes or apparatus for producing holograms
- G03H1/08—Synthesising holograms, i.e. holograms synthesized from objects or objects from holograms
- G03H1/0866—Digital holographic imaging, i.e. synthesizing holobjects from holograms
- G03H2001/0883—Reconstruction aspect, e.g. numerical focusing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Studio Devices (AREA)
- Automatic Focus Adjustment (AREA)
- Holo Graphy (AREA)
- Focusing (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention discloses a kind of for determining the auto focusing method of optimal focal plane.This method comprises: reconstruct (201) hologram image, and the first edge detection of (203) at least two reconstructed depths is executed based on the real part of reconstructed image, and the second edge detection in these reconstructed depths is executed based on the imaginary part of reconstructed image.This method further comprises: based on the statistic dispersion for being respectively relative to the first and second edge detections, to obtain first and second intelligibility measures of (204) each depth.This method further include: the focal plane of (205) at least one object is determined based on the comparison of the clarity scalar measurement of at least two depth, wherein the scalar measurement is based on the first and second intelligibility measures.
Description
Invention field
The present invention relates to the fields of digital hologram processing.More particularly it relates to which a kind of be used for digital hologram
The autofocus system and method for image procossing in art.
Background of invention
No lens holographic imaging can provide for the low cost solution that small object is imaged, because it is usually not
Need expensive and/or complicated optical module.When compared with the compact conventional microscopy for using lens, no lens holography at
As relatively large visual field can also be provided.In addition, holographic imaging allows good depth field imaging, so that large volume can pass through single
Image Acquisition is imaged.
However, in many cases, such as in the automaticly inspecting of object, between object of interest and visual detector
Distance be not it is known in advance, for example, the distance can be variable and can have significant random component.Such as this field
The digital hologram restructing algorithm (for example, the forward and backward using light field is propagated) known may usually need to provide such focal length and make
It is parameter to obtain high quality reconstruct.Since incorrect focusing can cause fuzzy image and to will cause specific application (all
As cell behavior analyze) difficulty, it is thus possible to expectation use a kind of method for automatically finding optimal focal plane and provide pair
The autofocus system answered.
It is known in the art that using including because becoming in the scalar image of the gradient magnitude of image coordinate, to determine nothing
The suitable focal plane of object of interest in lens holographic imaging.Such method can be based on by reconstructed wave amplitude angle value or correlative
The spatial gradient analysis for the holographic reconstruction image that (for example, scalar image intensity) is constituted.Such method is it is possible thereby to be characterized as being
Method based on amplitude.
For example, in entitled " the Detection of Waterborne Parasites of Onur Mudanyali et al.
In the paper of Using Field-Portable and Cost-Effective Lensfree Microscopy ", magnitude image
It is to be determined based on the image gradient amplitude because of change in two dimensional image coordinate, wherein image gradient passes through the water of reconstructed image
Gentle vertical Sobel operator convolution is come approximate.The variance of the magnitude image is used as focus measurement, wherein obtaining good acutance
Reach maximum value at reconstruct focal length with focus measurement in the case of contrast.
In entitled " the Fast Autofocus Algorithm of Mario A.Bueno-Ibarra et al.
In the paper of forAutomated Microscopes ", for the image obtained at different focal length by conventional micro-imaging
To determine that focus measures.The coke for the variance (SOB VAR) based on Sobel-Tenengrad gradient magnitude that the article disclose a kind of
Point measurement.The paper also discloses a kind of variance (LAP based on image Yu the absolute value of the convolution of discrete Laplace operator
VAR focus measurement).
Although such method based on amplitude is widely used, from the scalar image obtained by manipulating image derivative
The shortcomings that middle determining scalar focus measures is: the global search in entire interested depth bounds is carried out, to pass through
The maximization of scalar focus measurement determines optimal focal plane.Therefore, search range is reduced to enable to improve search speed
It will be advantageous.
Summary of the invention
The purpose of various embodiments of the present invention is to provide good and efficient auto focusing method and corresponding automatic focusing
System.
Object above is realized by method and apparatus according to the invention.
The advantages of various embodiments of the present invention, is not needing mechanical focus device.
The advantages of various embodiments of the present invention, is that off-line calculation image, which can be achieved, to be focused.
The advantages of various embodiments of the present invention, is, such as can get short image because may not be needed mechanical focus
Acquisition time.
The advantages of various embodiments of the present invention, is easy to quickly reduce focal length during automatic visual inspection
Search range.
The advantages of various embodiments of the present invention, is that the seldom iteration for only needing holographic image reconstruction is interested right to obtain
The good reconstructed image of elephant, for example, clearly, clearly with the image of well focussed.
In a first aspect, the present invention relates to one kind for determining (such as in reconstructed hologram image) at least one object
Focal plane (for example, optimal focal plane) auto focusing method (for example, computer implemented auto focusing method).The side
Method includes: the hologram image for reconstructing at least one object, in order to provide the reconstructed figure at multiple and different depths of focus
Picture.For example, the reconstructed image may include multiple reconstructed hologram images of two dimension, each reconstructed hologram image of two dimension is corresponding
In the different focal point depth that wherein the two-dimensional hologram picture is reconstructed.The reconstructed image includes for jointly encoding phase
The real component and imaginary component of position and amplitude information, for example, the reconstructed image is the phase and width before indicating focal plane medium wave
Spend the complex value image of both information.For example, the wavefront can correspond to form interference pattern and interacting with reference light wave
The object light wave of case, wherein these interference figures are recorded in original hologram picture, which is reconstructed with shape
At reconstructed hologram image.
The method further includes: first edge detection, example are executed to the real component at least two depth
Such as, the first edge dividually is executed in each of described at least two depth in the multiple different focal point depth
Detection, and detected at least two depth to execute second edge to the imaginary component, for example, dividually described
First edge detection is executed in each of at least two depth.
The method further includes: it is obtained based on the first statistic dispersion measurement detected relative to the first edge
First intelligibility measure of each of at least two depth, and based on second detected relative to the second edge
Statistic dispersion measurement obtains second intelligibility measure of each of described at least two depth.First statistic dispersion
Measurement and second statistic dispersion measurement can be respectively applied to the first edge testing result and the second edge inspection
The same mathematical operation for surveying result is corresponding.
The method further includes the comparison of: the clarity scalar measurement based at least two depth (for example, than
The value of clarity scalar measurement between at least two depth), to determine the focal plane (example of at least one object
Such as, optimal focal plane).The scalar measurement is based on first intelligibility measure and second intelligibility measure.
A kind of method of each embodiment according to the present invention can further comprise: in the mark reconstructed image at least
One object.It executes first edge detection and second edge detection, obtain first intelligibility measure and described
Second intelligibility measure and each step for determining the focal plane, can be additionally applied to correspond in the reconstructed image
The regional area of at least one described or each identified object.
In the method for each embodiment according to the present invention, the identification of steps can include: the digitlization reconstructed image,
Identify each connected region with same numbers value；And each of described each connected region of segmentation, for example, to be formed
The regional area of each identified object is used in the reconstructed image.
In the method for each embodiment according to the present invention, multiple objects can be identified, for example, at least one described object
It can be multiple objects.The method can further comprise: determine the multiple focal planes for corresponding to the multiple object.The side
Method can further comprise: the image-region for corresponding to each of the multiple object is spliced in corresponding focal plane one
It rises to form composograph, the composograph includes each of the multiple object of focus alignment.
In the method for each embodiment according to the present invention, the first statistic dispersion measurement and/or second statistics
Deviation measurement can be standard deviation.
Various embodiments of the present invention can provide a kind of auto focusing method for determining optimal focal plane as a result,.The method
Can include: reconstructed hologram picture identifies the object in the reconstructed image, and the real part based on the reconstructed image is deep one
First edge detection is executed for object at degree, the imaginary part based on the reconstructed pattern is directed to the object in the depth
Second edge detection is executed, is obtained based on the first standard deviation detected relative to the first edge in the depth place
The first readability of object is stated, and is obtained based on the second standard deviation detected relative to the second edge described
Second readability of object described in depth.The method can further comprise: be based on first standard deviation value and institute
The second standard deviation value is stated to determine the readability of the object described in the depth.
In the method for each embodiment according to the present invention, obtains the first intelligibility measure and/or obtain the second clarity
Measurement can include: obtain the gradient magnitude of the result of the first edge detection and/or second edge detection, and obtain
The standard deviation value of the gradient magnitude.The first intelligibility measure (for example, first readability) is obtained as a result, can include: is obtained
The first gradient amplitude of the result of the first edge detection, and obtain the first standard deviation of the first gradient amplitude
Value.Obtain the second readability can include: obtain the second gradient magnitude of the result of the second edge detection, and obtain institute
State the second standard deviation value of first gradient amplitude.
In the method for each embodiment according to the present invention, described at least two in the multiple different focal point depth are deep
Degree may include the depth being evenly distributed in preset range.
In the method for each embodiment according to the present invention, the depth being evenly distributed in the preset range can be wrapped
Include first quartile, the second quartile and the third quartile of the preset range.
A kind of method of each embodiment according to the present invention can further comprise: based on identified focal plane to determine
State the further depth of at least one of multiple and different depths of focus, and repeat the steps of: for this at least one into one
It walks depth and executes first edge detection and second edge detection, and obtain the of at least one further depth
One intelligibility measure and the second intelligibility measure.The method can further comprise: based on for it is described at least one further
The clarity scalar measurement determined by depth adjusts the focal plane.
A kind of method of each embodiment according to the present invention can further comprise: based on identified focal plane to determine
State the further depth of at least one of multiple and different depths of focus, be directed to based on the amplitude of the reconstructed image it is described extremely
A few further depth executes second stage edge detection, and is assessed based on the result of the second stage edge detection
The second stage intelligibility measure of the object.
In the method for each embodiment according to the present invention, at least two depth can be evenly distributed on the first pre- depthkeeping
It spends in range, and at least one described further depth may include the depth being evenly distributed in the second depth bounds, wherein
Second depth bounds are narrower than the first predetermined depth range.Second depth bounds can be by considering described at least two
The comparison of the clarity scalar measurement of a depth determines.
Thus various embodiments of the present invention can provide a kind of auto focusing method for determining optimal focal plane, the method packet
Include: reconstructed hologram picture identifies the object in the reconstructed image, and the real part based on the reconstructed image is in the first range
In predetermined depth set at for object execute first stage edge detection, the imaginary part based on the reconstructed image is first
First stage edge detection is executed for the object at the predetermined depth set in range, based on by the first stage
The standard deviation of each of the gradient magnitude of real and imaginary parts at edge of the edge detection object detected is commented
Estimate the first stage readability of the object at the predetermined depth set, and is based on the first stage readability
Mark the second range associated with the optimal focal plane in first range.
In each embodiment according to the present invention, the predetermined depth set may include being evenly distributed on first range
In several depth.In each embodiment according to the present invention, the predetermined depth set may include the of first range
One quartile, the second quartile and third quartile.
In each embodiment according to the present invention, the method can further comprise: the width based on the reconstructed image
Degree executes second stage edge detection for the object in second range, and is examined based on the second stage edge
The result of survey assesses the second stage readability of the object.
In the method for each embodiment according to the present invention, the first stage readability is assessed can include: obtain
The first gradient width of the result of each of predetermined depth place first stage edge detection associated with the real part
Value, and obtain the first standard deviation value of the first gradient amplitude.
In the method according to each embodiment, the first stage readability is assessed can include: obtain described predetermined
Second gradient magnitude of the result of each of depth place first stage edge detection associated with the imaginary part, and obtain
Obtain the second standard deviation value of second gradient magnitude.
In the method according to each embodiment, the method can further comprise: based on first standard deviation value and
Second standard deviation value determines the readability in each of the predetermined depth place object.
In the method according to each embodiment, execute first edge detection can include: by the real component or its
Partially (a part such as corresponding to identified object) carries out convolution (for example, convolution or application are discrete with Laplce's mask
Mathematics convolution algorithm).
In the method according to each embodiment, execute second edge detection can include: by the imaginary component or its
Part carries out convolution (for example, convolution or application discrete mathematics convolution algorithm) with Laplce's mask.
In second aspect, the present invention relates to a kind of for being used to determine the calculating equipment of focal plane in autofocus system,
The method for calculating equipment and being programmed to carry out each embodiment according to the first aspect of the invention.The calculating equipment can
Including memory, one or more processors and stores in the memory and be configured for by one or more
One or more programs that a processor executes.
The presently disclosed embodiments can provide in one kind such as autofocus system for determining the calculating of optimal focal plane
Equipment.The calculating equipment may include memory, one or more processors and storage in the memory and be configured
One or more programs for being executed by one or more of processors.One or more of programs may include for into
The following instruction operated of row: reconstructed hologram picture identifies the object in the reconstructed image, based on the reconstructed image
Real part executes first edge detection for object in a depth, and the imaginary part based on the reconstructed image is in the depth needle
Second edge detection is executed to the object, is obtained based on the first standard deviation detected relative to the first edge in institute
State the first readability of object described in depth, and based on the second standard deviation detected relative to the second edge come
Obtain the second readability of the object described in the depth.
The presently disclosed embodiments can provide in one kind such as autofocus system for determining the calculating of optimal focal plane
Equipment.The calculating equipment may include memory, one or more processors and storage in the memory and be configured
One or more programs for being executed by one or more of processors.One or more of programs may include for into
The following instruction operated of row: reconstructed hologram picture identifies the object in the reconstructed image, based on the reconstructed image
Real part executes first stage edge detection for object at the predetermined depth set in the first range, is based on the reconstructed figure
The imaginary part of picture executes first stage edge detection for the object at the predetermined depth set in first range,
Standard deviation based on each of real and imaginary parts by the first stage edge detection target edges detected come
The first stage readability of the object at the predetermined depth set is assessed, and is based on the first stage clear journey
Degree identifies the second range associated with the optimal focal plane in first range.
Various embodiments of the present invention may also refer to a kind of autofocus system, and the autofocus system includes being used for direction
The light source of sample radiation light in test and the imager of the hologram for acquiring the sample.The autofocus system can be into
One step includes a kind of calculating equipment for being used to determine focal plane of each embodiment according to the present invention.
In the third aspect, the invention further relates to a kind of computer program product, the computer program product is used to work as
According to the first aspect of the invention each is executed when executing in the calculating equipment of each embodiment according to the second aspect of the invention
The method of embodiment.
Special and preferred aspect of the invention illustrates in appended independence and dependent claims.In dependent claims
Feature can be combined as suitable in the technical characteristic of the feature of independent claims and other dependent claims, and not only
It is as it is clearly illustrated in the claims.
Will become obvious from (all) embodiments described below in terms of these and other of the invention and
It will be illustrated with reference to these embodiments.
Brief description
Fig. 1 shows the schematic diagram of the autofocus system of each embodiment according to the present invention.
Fig. 2 shows the flow charts of the method for determining optimal focal plane of explanation each embodiment according to the present invention.
Fig. 3 shows the stream for explaining the method for the optimal focal plane for determining object of each embodiment according to the present invention
Cheng Tu.
Fig. 4 shows the flow chart for explaining the object segmentation methods of each embodiment according to the present invention.
Fig. 5 illustrates the edge detection of each embodiment according to the present invention and the method for marginal definition assessment.
The method that Fig. 6 illustrates the readability of the determination object of each embodiment according to the present invention.
The method that Fig. 7 illustrates the optimal focal plane for determining one or more objects of each embodiment according to the present invention.
The method that Fig. 8 illustrates the optimal focal plane for determining one or more objects of each embodiment according to the present invention.
Fig. 9 to Figure 19 schematically illustrates the method for determining optimal focal plane of each embodiment according to the present invention
Each stage.
Figure 20 illustrate according in this field may known to method because become in the readability of different focal point depth.
Figure 21 show as can each embodiment through the invention obtain because becoming in the readability of different focal point depth
With the aspect for explaining various embodiments of the present invention.
These attached drawings only schematically and not restrictive.It in the accompanying drawings, can be by some elements for illustrative purpose
Size is amplified and is not drawn on scale.
Any appended drawing reference in claim is not necessarily to be construed as limitation range.
In different drawings, identical appended drawing reference refers to the same or similar element.
The detailed description of illustrative embodiments
Although will about specific embodiment and with reference to certain figures description the present invention, but the invention is not restricted to this and only by
Claim limits.Described attached drawing only schematically and not restrictive.It in the accompanying drawings, will for illustrative purpose
The size of some elements is amplified and is not drawn on scale.Size and relative size do not correspond to practice practical contracting of the invention simultaneously
Subtract.
In addition, term " first " in the description and in the claims, " second " etc. are in similar element
Between distinguish, and not necessarily for provisionally, spatially, with sequence or description order in any other manner.It should
Understand, these terms so used can be interchanged under proper environment, and the embodiment of invention described herein can
It is sequentially operated with other other than being described herein or explaining.
In addition, term " top " in the description and in the claims, " bottom " etc. are for descriptive purposes
And not necessarily for description relative position.It should be understood that these terms so used can be interchanged under proper environment, and
And the embodiment of invention described herein can sequentially be operated with other other than being described herein or showing.
It should be noted that term " includes " used in claim should not be construed as limited to the device hereafter listed；
It is not excluded for other elements or step.Thus it should be interpreted to specify has stated feature, integer, such as institute's appellation
Step or component, but do not preclude the presence or addition of other one or more features, integer, step or component or its group.
Therefore, a kind of range of word " equipment including device A and B " should not be defined to the equipment being only made of component A and B.
This means that unique component related to the present invention of the equipment is A and B.
The spy for combining the embodiment to describe is meaned to the reference of " one embodiment " or " embodiment " in this specification
Determine feature, structure or characteristic is included at least one embodiment of the invention.As a result, the phrase " in one embodiment " or
" in embodiment " different establish a capital of appearance in each place through this specification quotes identical embodiment, but can be as
This.In addition, in one or more embodiments, specific features, structure or characteristic can combine in any suitable manner, such as
Those of ordinary skill in the art will be apparent according to the disclosure.
Similarly, it should also be appreciated that in the description of exemplary embodiment of the present invention, for the streamlining disclosure and auxiliary
Help the purpose to the understanding in terms of one or more inventions in terms of each invention, each feature of the invention sometimes by
It is grouped into single embodiment, attached drawing or its description together.However, the disclosure method is not construed as required by reflection
The invention of protection needs the intention of the more features than being expressly recited in each claim.On the contrary, such as the appended claims
Reflected, is present in the feature fewer than all features of single previously disclosed embodiment in terms of invention.Therefore, it retouches in detail
Thus claim after stating is expressly incorporated into the detailed description, wherein each single item claim itself represents Ben Fa
Bright separate embodiments.
In addition, while characterized as some embodiments include some features included in other embodiments but without it
Other features for including in his embodiment, but the combination intention of the feature of different embodiments is fallen within the scope of the present invention, and
And form different embodiments as understood by those skilled in the art.For example, in the appended claims, it is claimed
Embodiment in any embodiment can be come with any combination using.
In description provided by herein, a large amount of details are set forth.It is, however, to be understood that can there is no these
The embodiment of the present invention is practiced in the case where detail.In other instances, it is not shown specifically well-known method, structure
And technology, in order to avoid obscure understanding of the description.
In a first aspect, the present invention relates to a kind of at least one objects in for example reconstructed hologram image of determination
The auto focusing method (for example, computer implemented auto focusing method) of focal plane (for example, optimal focal plane).This method packet
It includes: reconstructing the hologram image of at least one object, in order to provide the reconstructed image at multiple and different depths of focus.
This method further comprises: holding at least two depth in the multiple different focal point depth to real component
The detection of row first edge, for example, first edge detection is dividually executed in each of at least two depth, and
Second edge detection is executed to imaginary component at least two depth, for example, dividually every at least two depth
First edge detection is executed in one.
This method further comprises: obtaining at least two based on the first statistic dispersion measurement detected relative to first edge
First intelligibility measure of each of a depth, and based on the second statistic dispersion measurement detected relative to second edge
To obtain second intelligibility measure of each of at least two depth.The measurement of first statistic dispersion and the second statistic dispersion are surveyed
Amount can be corresponding with the same mathematical operation for being respectively applied to first edge testing result and second edge testing result.In basis
In the method for various embodiments of the present invention, the measurement of the first statistic dispersion and/or the second statistic dispersion measurement can be standard deviation.
This method further comprises: the comparison of the clarity scalar measurement based at least two depth is (for example, compare
The value of clarity scalar measurement between at least two depth), to determine the focal plane of at least one object (for example, most
Excellent focal plane).The scalar measurement is based on the first intelligibility measure and the second intelligibility measure.
This method can further comprise: identify the object in reconstructed image, for example, identify it is each right in reconstructed image
As.The identification of steps can include: digitize reconstructed image, identify each connected region with same numbers value；And point
Each of each connected region is cut, for example, to form in reconstructed image the partial zones for being used for each identified object
Domain.
In the method for each embodiment according to the present invention, obtains the first intelligibility measure and/or obtain the second clarity
Measurement can include: obtain the gradient magnitude of the result of first edge detection and/or second edge detection, and obtain the gradient width
The standard deviation value of value.
A kind of method of each embodiment according to the present invention can further comprise: be determined based on identified focal plane more
The further depth of at least one of a different focal point depth, and repeat the steps of: for this, at least one is further deep
Degree executes first edge detection and second edge detection, and obtains the first intelligibility measure of at least one further depth
With the second intelligibility measure.This method can further comprise: based on the clear scale for being directed at least one further depth
Measurement adjusts focal plane.
A kind of method of each embodiment according to the present invention can further comprise: be determined based on identified focal plane more
The further depth of at least one of a different focal point depth is directed at least one based on the amplitude of reconstructed image further
Depth executes second stage edge detection, and assessed based on the result of second stage edge detection object second stage it is clear
Clear degree measurement.
In the method for each embodiment according to the present invention, at least two depth can be evenly distributed on the first predetermined depth model
In enclosing, and at least one further depth may include the depth being evenly distributed in the second depth bounds, wherein the second depth
Range is narrower than the first predetermined depth range.Second depth bounds can pass through the clarity scalar measurement of at least two depth of consideration
Compare to determine.
In each embodiment according to the present invention, predetermined depth set may include be evenly distributed on it is several in the first range
Depth.In each embodiment according to the present invention, predetermined depth set may include the first quartile of the first range, the two or four
Quantile and third quartile.
In each embodiment according to the present invention, this method can further comprise: the amplitude based on reconstructed image,
Second stage edge detection is executed for object in two ranges, and object is assessed based on the result of second stage edge detection
Second stage readability.
In the method for each embodiment according to the present invention, first stage readability is assessed can include: obtain predetermined
The first gradient amplitude of the result of each of depth place first stage edge detection associated with real part, and obtain the
First standard deviation value of one gradient magnitude.
In the method according to each embodiment, first stage readability is assessed can include: obtain in predetermined depth
Second gradient magnitude of the result of first stage edge detection associated with imaginary part at each, and obtain the second gradient width
Second standard deviation value of value.In the method according to each embodiment, this method can further comprise: be based on the first standard deviation
Value and the second standard deviation value determine the readability in each of predetermined depth place object.
Fig. 2 shows explanation each embodiments according to the present invention for determining focal plane (for example, limiting coke for determining
The depth of focus parameter of plane) auto focusing method flow chart.It can be computer according to the method for each embodiment to realize
Method, for example, for the side that is executed on processing equipment (for example, such as below in conjunction with processing equipment 20 described in Fig. 1)
Method.The focal plane can be the optimal focal plane of object in reconstructed hologram image, for example, in the case where considering the focal plane
It can carry out reconstructed hologram picture using digital hologram restructing algorithm, so that the reconstructed image of high quality is generated.' most
It is excellent ' it can be referred to the focal plane essentially corresponded at a distance from object of interest to imaging plane.' optimal ' can be referred to pass through into
The value that the algorithm optimization of this measurement or quality factor (for example, clarity scalar measurement mentioned below) obtains, characterization is through weight
At least one attribute of the instruction picture quality of composition picture, such as acutance and/or clarity.In addition to clarity and/or acutance it
Outside, the cost measuring or quality factor are also conceivable to other image quality measurements, such as contrast, signal-to-noise ratio and/or entropy and/
Or information theory measurement.' optimal ' can refer only to obtain by algorithm optimization process as a result, not necessarily implying that being obtained
Obtain the subjective understanding of result.Technical staff is further appreciated that, due to for example to the number of iterations, processing time or predetermined tolerance limit model
The limitation enclosed, such algorithm optimization can suspend at the value for being substantial access to theoretially optimum value.In addition, such algorithm optimization is preferable
Ground provides global optimum, but in each embodiment according to the present invention, algorithm optimization can also provide local optimum, for example, mesh
The local maximum or local minimum of scalar functions.
The method of each embodiment according to the present invention can include: receiving includes sample (for example, including at least one object
Sample) optical information hologram image.For example, such hologram image can be as input be received, for example by being imaged
The original hologram picture that device directly acquires.
The method of each embodiment according to the present invention is the following steps are included: reconstruct the hologram image of 201 at least one object
201, in order to provide the reconstructed image at multiple and different depths of focus.For example, reconstructed image may include multiple two dimension warps
Reconstructed hologram picture, each reconstructed hologram image of two dimension correspond to the different focal point depth that the two-dimension holographic image is reconstructed.
Thus reconstructed image at multiple and different depths of focus can form the two dimensional image of 3-D image or different focal point depth
It stacks.When hereafter reference depth or the depth of focus, reference hologram is (also referred to as burnt flat as acquisition plane and reconstruction plane
The distance between face).However, technical staff is it should be clear that this is only that focal plane facilitates parametrization, and therefore, ' depth '
Or ' depth of focus ' should not be construed as limited to such parametrization, and may include reconstructing on it for limiting in space
One or more parameters on the surface of hologram image.It is executed wherein in space for example, depth or the depth of focus can be referred to limit
Any combination of the parameter of the plane of holographic reconstruction or even non-planar surfaces.Therefore, reference depth or coke in this description
In the case where point depth, it should be understood that this may refer to limit in space, and execution is related to the depth or the depth of focus wherein
At least one parameter on the surface (for example, plane) of the holographic reconstruction of connection.
Reconstructed image includes real component and the imaginary component for jointly encoding phase and amplitude information.For example, through weight
Composition picture includes phase information and amplitude information, it may for example comprise real and imaginary parts.For example, for each reconstructed image lattice
Position (k, l), reconstructed image may include complex value ckl=akl+bkl.i (wherein i indicates imaginary unit), it can combine
Ground encodes both phase information and amplitude information of wavefront.Reconstructed image can be the phase before indicating focal plane medium wave as a result,
The complex value image of position and both amplitude informations.For example, the wavefront can correspond to form and interacting with reference light wave
The object light wave of interference figure, wherein these interference figures are recorded in original hologram picture, which is weighed
Structure is to form reconstructed hologram image.For example, reconstructed image may include phase information and amplitude information, for example, real part and
Imaginary part.
Thus reconstructed image includes phase information and amplitude information.As known to technicians, amplitude and phase letter
Breath may be expressed as the form with the complex field (for example, the array for indicating complex field) of real component and imaginary component.It the domain can example
Such as limited by the cartesian coordinate in image reconstruction plane.However, various embodiments of the present invention are without being limited thereto, because of ability
It can mathematically facilitate expression that field technique personnel, which will be understood that using plural number,.However, such complex representation can for example exist
Be advantageously used in various embodiments of the present invention in following meaning: imaginary component image and real component image (for example, include at the same time
When at least partly complementary information) the relevant information of amplitude and the relevant both information of phase can be advantageously comprised without superfluous
It is remaining.
Reconstructed hologram picture may include the multiple reconstructed images or reconstructed hologram that reconstruct corresponds to multiple focal planes
As may include iterative algorithm, wherein corresponding to focal length at least one at each step determines at least one reconstructed image,
At least one correspondence focal length is by reducing search range (for example, proceeding in depth resolution method from roughness depth grade
Fine depth level) it determines.
Referring to Fig. 2, in operation 201, thus the hologram image of the restructural optical information including sample is provided reconstructed
Image.Hologram image can be provided by imaging system (all imaging systems 10 explained as described below and referring to Fig.1).In addition,
Sample may include one or more objects at different depth (for example, in different focal planes).These different depth
And/or different focal planes may be unknown in advance, (is such as led to by container or flowing for example, it may be possible to only know in advance and be included in
Defined by the boundary in road) in pre- broad range, and can each embodiment according to the present invention determine.
The advantages of digital holography is that the interference figure between reference beam and object beam can be captured, the wherein interference
Pattern includes the information of the three-dimensional structure about imaging volume.Reconstruct can be for example realized by angular spectrum method or convolution method
201 hologram images.Such method or the like is generally well-known in the art and is no longer discussed in detail.
Digital holography uses reconfiguration technique as known in the art, is advantageously carried out the amplitude and phase information to wavefront
The reconstruct of the two.It is known in the art that the reconstructed hologram picture (example at the focal length for being supplied to restructing algorithm as parameter
Such as, the reconstructed image of two dimension).Due to original hologram picture may include detailed object wavefront information (e.g., including phase letter
Breath), therefore the reconstructed image of object can be determined in any focal plane by appropriately changing focal length parameter.Routinely aobvious
In the case where micro- art, automatic focusing can be realized until obtaining the image focused by mechanically changing focal length, it can from single original
Beginning hologram image calculates multiple images plane.
Different from original hologram picture (it may include the interference figure for being not easy to explain by visual inspection), reconstructed figure
As can be the image that can be used for direct vision inspection, for example, the physical space geometry of object of interest can be indicated directly.So
And the optimal depth of reconstructed objects in images or optimal focal plane may be still to be determined.Such as by using identical or different
Restructing algorithm (such as using another restructing algorithm that is higher but being capable of providing higher reconstruction quality is required in calculating) it is right
Reconstruct is iterated, and such focal plane of one or more object of interest can be used for improving the quality of reconstructed image.One
Or the focal plane of multiple object of interest can be also used for index and be directed to different images region (for example, for different objects) no
The same depth of focus, so that single two dimensional image can be folded by corresponding to the stacking of the holographic reconstruction image of different depth,
With for for example by pattern identification, machine learning, measurement and/or other characterize algorithms come carry out easy visual inspection and/
Or it is further processed.
This method is further can include: the object in 202 reconstructed images of mark, for example, identifying in reconstructed image
Each object.For example, can be based on the real part of reconstructed image, in a depth for object (for example, in identified object
Each) execute the further step that first edge detects.For example, can be based on the imaginary part of reconstructed image, in the depth needle
Second edge detection is executed to object (for example, for each of identified object).The first and second edges are executed as a result,
Detection obtains the first and second intelligibility measures and can be applied in reconstructed image correspond to the step of determining focal plane
The regional area of at least one identified object or at least one each identified object.In this class identification operation 202, warp
One or more objects in reconstructed image can be identified and separated from one another.Object Segmentation can advantageously facilitate detection one or more
The optimal focal plane of each of a object of interest, for example, the separated inspection of the optimal focal plane for each corresponding objects
It surveys.The exemplary details of the object in 202 reconstructed images of mark are further discussed (for example, Object Segmentation below in reference to Fig. 4
Exemplary details).
It can get single original hologram picture as a result, detect multiple objects from the original hologram picture.For each detection
The object arrived, can each embodiment according to the present invention determine focal plane.The information can for example be used to that multiple objects will to be corresponded to
Image-region is stitched together, and each image-region is reconstructed in corresponding focal plane.By this method, composite diagram can be created
Picture, the composograph include each of multiple objects of focus alignment.
This method further comprises: holding at least two depth in the multiple different focal point depth to real component
The detection of 203 first edge of row, for example, first edge detection is dividually executed in each of at least two depth, and
And detected at least two depth to execute second edge to imaginary component, for example, dividually every at least two depth
First edge detection is executed in one.Edge detection can refer to shrilly change for brightness of image in reference numbers image with/
Or the image processing method with discrete point.Other than enabling such marginal information to be used in the form of edge image, side
Edge detection does not need to imply the picture position for determining and forming such edge.This method as a result, can include: the edge of test object or
Boundary.The detection edge or boundary include: to execute first edge detection to the real part of reconstructed image, for example, for detecting
The edge of one depth object, and second edge detection is executed to the imaginary part of reconstructed image, for example, for detecting in the depth
The edge of object at degree.
The detection 203 at the edge or boundary of object (for example, any object in identified object) can include: in predetermined model
(for example, in the first depth Z in enclosingaWith the second depth ZbBetween) in different depth at focal plane global search.As a result,
At least two depth may include multiple depth in the preset range, for example, being evenly spaced apart in this range.Multiple depth
Possibility candidate's depth of the focal plane (for example, optimal focal plane) of corresponding objects can be construed as limiting.
In each embodiment according to the present invention, the depth in preset range can begin to the detection 203 at edge or boundary
Spend Zj, for example, in ZaWith ZbBetween depth Zj.Start depth ZjThe vision to reconstructed image for example for example can be passed through by user
It checks to determine.Then, it completes in depth ZjIt, can be away from beginning depth Z after the edge detection at placejRegular section it is next
Depth gradually implements global search for edge detection.For example, starting depth can be 10000 microns (μm), for example, from
The distance that 10000 μm of imager, and next depth is 10100 μm or 9900 μm, away from the section for starting 100 μm of depth.
In each embodiment according to the present invention, it can be examined in convolution algorithm using Laplce's mask in order to edge
It surveys, for example, edgeimage=conv (image, edgeoperator)), or it is alternatively represented as edgeimage=image*
edgeoperator, wherein * refers to convolution operator, for example, discrete picture convolution algorithm.In these formula, ' image ' is indicated
The matrix of reconstructed image at given depth, and ' edgeoperator' serve as the mask or operator of convolution algorithm.
However, various embodiments of the present invention are not limited to Laplce's mask, and the convolution algorithm can also relate to this
Known difference edge detection filter in field, for example, the width for calculating Sobel gradient, Sobel-Tenengrad gradient
The filter of value, or the higher derivative filter for generating scalar edge image or scalar edge enhancement type image.It is rolling up
After product, the resulting matrix ' edge for indicating target edges can getimage’。
In each embodiment according to the present invention, the target edges of the real and imaginary parts relative to reconstructed image can be distinguished
It is denoted as ' edgeimage,real' and ' edgeimage,imaginary', and can determine as follows:
edgeimage,real=conv (real_image, edgeoperator), and
edgeimage,imaginary=conv (imaginary_image, edgeoperator),
Wherein, edgeoperatorIndicate edge detection filter, such as gradient magnitude filter or Laplce's filtering
Device or another suitable edge detection convolution filter as known in the art.
In the method for each embodiment according to the present invention, thus first edge detection is executed to real component can include: answer
With the first edge Fault detection filter for detecting edge, and second edge detection is executed to imaginary component can include: be used for
Detect the second edge Fault detection filter at edge.
This method further comprises: obtained based on the first statistic dispersion measurement detected relative to first edge 204 to
First intelligibility measure of each of few two depth, and based on the second statistic dispersion detected relative to second edge
It measures to obtain second intelligibility measure of each of at least two depth.First statistic dispersion measurement and second count from
Difference measurements can be corresponding with the same mathematical operation for being respectively applied to first edge testing result and second edge testing result.
The method of each embodiment according to the present invention further comprises next step below: being determined 204 identified
Each of different depth place target edges or the readability on boundary.The determination 204 includes: based on relative to first edge
First standard deviation of detection obtains the first intelligibility measure (for example, readability) in the depth object, and is based on
Relative to the second standard deviation of second edge detection, the second intelligibility measure in the depth object is obtained (for example, clear
Degree).
In some embodiments, it is denoted as edgeclearnessIntelligibility measure (for example, readability) can based on behaviour
Make the statistical result of edge detection in 203 to determine, such as:
edgeclearness=std (gradientmagnitude (edgeimage)), or alternatively indicate,
edgeclearness=std (| gradient (edgeimage) |), or in another replacement formula,
Wherein, edgeimageIt is the gained matrix obtained in operation 203, gradient (M) indicates the gradient of M (in this hair
In bright embodiment, M is gained matrix edgeimage), gradientmagnitude (M) indicates the amplitude of the gradient of M, such as
European norm, and std (N) indicates the standard deviation (in an embodiment of the present invention, the gradient that N is edgeimage) of N.
As defined by equation above, the edge edge in each of different depth place object can getclearness
(it is scalar real number value).In addition, each embodiment according to the present invention, relative to reconstructed objects in images edge real part and
The readability of imaginary part can be dividually determined, and be denoted as edge respectivelyclearness,realWith
edgeclearness,imaginary.These values can be determined as follows by equation above:
edgeclearness,real=std (| conv (edgeimage,real,gradientoperation) |), and
edgeclearness,imaginary=std (| conv (edgeimage,imaginary,gradientoperation)|)。
This method further comprises: the comparison of the clarity scalar measurement based at least two depth is (for example, compare
The value of clarity scalar measurement between at least two depth), to determine the focal plane of at least one object (for example, most
Excellent focal plane).The scalar measurement is based on the first intelligibility measure and the second intelligibility measure.
The determination focal plane can include: exporting at least one indicates the value of focal plane.For example, determining focal plane can include:
Generation 205 includes the image of the information on the optimal focal plane about the or each object.
The focal plane (for example, focal length) that each embodiment according to the present invention obtains can provide the well focussed of image, and
Therefore the clear and/or sharp edges of objects in images be can provide.Such clear and/or sharp edges can be by big gradient value
(for example, being greater than gradient value in similar out focus (out-of-focus) reconstruct) and big readability characterize.
Each embodiment according to the present invention can assess the survey of clarity scalar to determine the optimal focal plane of object as follows
Amount (for example, measurement EC):
However, such scalar measurement may also include another mark applied to the first intelligibility measure and the second intelligibility measure
Amount summarizes operation, such as absolute value and quadratic sum, the maximum value of absolute value, or in general, be suitable as mathematical meaning
On measurement norm or even semi-norm any function.
It as a result, can standard deviation and edge image imaginary part for example based on the edge image real part in each depth object
Standard deviation, to calculate the clarity scalar measurement in each of different depth place object, as described above.
By comparing the value of the clarity scalar measurement (for example, measurement EC) at different depth, the depth with maximum EC value
Degree can be identified as ZaWith ZbBetween preset range in object optimal depth or optimal focal plane.Then, 205 packets are produced
Include the image of the information on the optimal focal plane about object.
Figure 20 and Figure 21 shows the figure for explaining the readability at different depth.Referring to Figure 20, this field is illustrated
Intelligibility measure known to middle possibility.The clear journey that y-axis indicates the standard deviation (std) by obtaining magnitude image to determine
Degree, can be briefly expressed as applied statistics and summarize operation, for example, to the complex values of hologram image are reduced to it is corresponding
The edge-detected image obtained after plural modulus value calculates standard deviation, for example, as can be by operation performed by being used to indicate
The operation template of sequenceSummarized.
Referring to Figure 21, y-axis indicates intelligibility measure determined by each embodiment according to the present invention (for example, as fixed above
The measurement EC of justice), it can be briefly expressed as complex value intelligibility measure being reduced to its modulus of complex number, complex value intelligibility measure
Real component and imaginary component by the statistics derived from complex value edge image summarize operation (for example, standard deviation) to calculate,
For example, such as can be by being used to indicate the operation template of the sequence of performed operationSummarized.
, it is surprising that it is more regular that curve shown in curve ratio Figure 20 in Figure 21 can be observed, for example, show compared with
Unconspicuous local extremum.Such regular curve (for example, smoother curve) is convenient for determining that there may be optimal focal planes
" hot spot region ", as will be referring to Fig. 9 is discussed in detail to Figure 19.It is searched as a result, search process can be simplified and therefore reduced
The rope time.For example, reducing the risk that optimization process falls into local maximum (for example, corresponding to suboptimum focal plane).In addition, compared with
Smooth optimisation criteria function can make it possible for more efficient searching algorithm, such as, be based at least partially on excellent
Change function relative to the gradient of Optimal Parameters (for example, relative at least one of the depth of focus or the restriction focal plane to be determined
The derivative of parameter) optimization method.
Fig. 3 shows the another exemplary side for explaining the optimal focal plane of determination object of each embodiment according to the present invention
The flow chart of method.The illustrative methods explained in Fig. 2 can provide global search, and the method in Fig. 3 can be predetermined by assessing
Depth set (less than those depth set required for global search) carrys out acceleration search process.In addition, according to institute in such as Fig. 3
The method of explanation, each embodiment according to the present invention can carry out rough search and fine search to determine focal plane (for example, most
Excellent focal plane).
Referring to Fig. 3, reconstruct 201 hologram images and optionally identify in 202 reconstructed images object (for example,
Divide the object in hologram image) after, it, can be at predetermined depth set (for example, in Z in operation 303aWith ZbBetween
In one preset range) for object execute first stage edge detection, wherein real part of the edge detection based on reconstructed image
And imaginary part.The first stage edge detection may include relative to the first edge detection of real part and relative to the second edge of imaginary part
Detection, to for example respectively obtain edge according to equation provided aboveimage,realAnd edgeimage,imaginary。
In each embodiment according to the present invention, the predetermined depth set in the first range may include being evenly distributed on ZaWith
ZbBetween three depth, will be further discussed referring to Fig. 9 to Figure 19.However, sampled depth (for example, predetermined depth)
Number do not need to be limited to three points.
The method of each embodiment according to the present invention can further comprise: assessment 304 is at each of predetermined depth place
The first stage intelligibility measure of object.First stage readability assessment can include: provided above right for example, by application
Answer the edg in equationeimage,realAnd edgeimage,imaginaryTo determine edgeclearness,realAnd edgeclearness,imaginary, and
And then by calculating component edgeclearness,realAnd edgeclearness,imaginaryScalar norm determine the first measurement
EC1.As a result, the readability in each of predetermined depth place object is based on the object at each predetermined depth
The standard deviation of the real part at edge and the standard deviation of imaginary part calculate.
Based on the measurement of (for example, three equally distributed depth in the first range) first EC1 at predetermined depth
Value, can construct EC1 curve by connecting these positions EC1.Such EC1 curve includes about the rule explained in such as Figure 21
The information of curve.In addition, gradient of the EC1 curve in the first range can reveal that the global peak (example about regular curve
Such as, correspond to optimal focal plane) information.For example, if EC1 curve is intended to rise, it is contemplated that optimal focal plane is fallen in
In the upper section of first range.In addition, if EC1 curve is intended to decline, then it is contemplated that optimal focal plane falls in the first model
In the relatively lower part enclosed.By utilize gradient, can determine be less than and fall in it is in the first range, between Za' and Zb'
Second range.
Then, second stage edge detection 305 can be executed for object at the different depth in the second range.In root
According in various embodiments of the present invention, the amplitude of reconstructed image can be used in second stage edge detection, such as by
edgeimage,amplitude=conv (amplitude_image, edgeoperator)It is expressed, wherein amplitude_image table
Show the matrix of the magnitude image of reconstructed image at given depth.The magnitude image can correspond to the complex component of reconstructed image
The modulus of complex number of real_image and imaginary_image.
In some embodiments of the invention, second stage can be globally executed at different depth in the second range
Edge detection.It, can the less depth required for than the global search in the second range in each embodiment according to the present invention
It spends and advantageously executes second stage edge detection at set.
It can further comprise according to the method for each embodiment: be determined based on the result for executing second stage edge detection
The second stage readability of 306 objects.Second stage readability assessment can include: by by edgeimage,amplitudeUsing
The second measurement EC2 is determined in following formula:
By comparing the value of EC2 associated with the different depth in the second range, the depth with maximum EC2 value can quilt
It is identified as Za'With Zb' between the second range in object optimal depth or optimal focal plane.Then, it in operation 205, generates
Image including the information on the optimal focal plane about object.
It will be apparent to those skilled in the art that in general such regulation can appoint in three-stage process, four phase process or
What realized in the stage of number, wherein previous stage optimal focal plane obtained is at least partially defining in next stage
Depth range search.Those skilled in the art be also apparent that reconstructed hologram as the step of can execute online, for example, working as
When needing corresponding depth in a single stage, specific focal plane is only reconstructed.
In each embodiment such as explained by Fig. 3, (for example, Z in wider rangeaWith ZbBetween the first range)
The operation 303 and 304 of execution may make up the rough search of this method, and (for example, Z in narrower rangea'With Zb'Between
Two ranges) execute operation 305 and 306 may make up this method the optimal focal plane to object fine search.Although each
A rough search stage and a fine search stage, but more than one coarse filtration search phase and one are discussed in embodiment
The above fine search stage also falls within the scope of the present invention.One advantage of each embodiment is efficiently and with low to be calculated as
This (such as, it can be achieved that low processing time) executes search process.
However, each embodiment according to the present invention, it should be clear that the visually interpretable curve of construction is not necessary,
As described above.Furthermore, it is possible to appropriately reduce search range using other numerical optimizations as known in the art.
For example, predetermined depth range (for example, in the first phase) may include two boundary points (for example, ZaAnd Zb) and
(for example, being included in range [Za,Zb] in) at least two internal points.Can then it refuse near with minimum EC1 value
The boundary point of internal point, and this method can advance to next step, the next step have boundary point by not refusing and
The reduced depth range search that the internal point with minimum EC1 value is formed.As it is known in the art, such directly search
The advantages of rope algorithm, is only to need to calculate an additional internal point in the next steps.In addition, as it is known in the art,
It can be searched for using the alternative for dividing the region of search, such as by application gold section.
In addition, other searching methods as known in the art, such as gradient decline, Newton method, quasi-Newton method can be applied.Its
He is also known linear search method in the art, can also be using these methods without creative effort.Alternatively,
Method can be searched for using trust region.Various embodiments of the present invention are also not necessarily limited to linear search, but can be for example including grid
Search or the optimization of other vector parameters, because ' depth ' can cover vector value parameter in the context of the disclosure, for example, not
It only limits to the distance of imaging plane, and limits the gradient relative to the imaging plane.
In the method for each embodiment according to the present invention, it can be iteratively applied searching method, be substantial access to until reaching
The depth of the maximum value of objective function (for example, EC1 and/or EC2) or other the one or more correlation optimizations for limiting focal plane
Parameter.' being substantial access to ' can be for example corresponding to the predetermined tolerance limit range of depth, the predetermined tolerance limit range of target function value, scheduled
It is the number of iterations, quality factor, as known in the art for another stopping criterion of numerical optimization or above-mentioned any group
It closes.However, as discussed above, such searching method can be applied to the first stage (for example, in rough grade) iteratively to contract
Small search range, the search range can further reduce in second stage (for example, being in fine-grade as discussed above).Example
Such as, second stage can execute global search in the range such as reduced by the first stage.In addition, second stage can be used such as by having
There is objective function determined by the modulus of complex number of the plural number of real component and imaginary component, the real component and imaginary component correspond to be applied respectively
In the collect statistics of the edge image obtained from real number and imaginary number holographic reconstruction image, for example, EC1 as described above.However,
Second stage also can be used as by being applied to from the edge image (for example, magnitude image) that modulus of complex number holographic reconstruction image obtains
Objective function determined by collect statistics, all exemplary scalar measurement EC2 as described above.
Fig. 4 shows the method for identifying the object in reconstructed image for explaining each embodiment according to the present invention
The flow chart of (for example, object segmentation methods).For identifying this class object of object or as known in the art for identifying
The another method of object can be applied independently for corresponding with each reconstructed depth each in the various embodiments of the invention
On reconstructed 2D image, for example, object can be in each two dimension in the stacking of the reconstructed image formed by multiple reconstruct depth
It is identified in image.
For identifying this method of the object in reconstructed image can include: 401 reconstructed images of pretreatment are to improve it
Quality.For example, executable filter to reduce the noise in acquired image, for example, in each embodiment according to the present invention, it can
With only application filtering, or filtering operation can be applied in conjunction with other noise reduction techniques.Executable self-adaptive harmonics detection
To improve the image sharpness and/or contrast of acquired image.In addition, both minimum filtering and self-adaptive harmonics detection can answer
For in an embodiment.However, image preprocessing 401 can be considered optional.
In next operation, reconstructed image 402 can for example, by each pixel assign adaptation value digitized, from
And obtain digitized reconstructed image.Herein, " digitlization " can refer in particular to generation to the binary of reconstructed image pixel value
Change.For example, binary value " 1 " can be assigned to the pixel if the gray scale of pixel is greater than or is greater than or equal to threshold value, and
If the gray scale of pixel falls below threshold value, another binary value " 0 " can be assigned to the pixel.As a result, can get warp
The binary picture of reconstructed image.Such as it can be in the case where considering image intensity (for example, passing through application image intensity normalizing
Change) threshold value.
This method further may include 403 connection of identification or continuum, for example, being marked in digitized reconstructed image
It is denoted as the region of binary one.Each connected region can be grouped into and number.For example, the pixel in each connected region can
The group identified by unique identifying number is formed to be grouped into.Each group can be considered as potential object.For example, can be based on neighbour
Recency measurement or morphologic criteria by multiple connected regions organize into groups together, if for example, so as to multiple connected regions close to or
Only separately lower than the small size of the binary quantization threshold value referred in step 402 (for example, being marked in digitized reconstructed image
It is denoted as the small size of " 0 "), then they are attributed to identical object.Alternatively, if it is desired, for example, if picture contrast is sufficient
To allow the high-fidelity digital of reconstructed image, then each connected region can be identified as corresponding group, for example, without will be more
A region is collected into single group.
This method can further comprise: if the group (for example, group #1) of connected region, which has, is greater than or equal to threshold value
Size, then by the group identification 405 be object.However, if the group (for example, group #2) of connected region, which has, is less than threshold
The size of value, then the group can be considered impurity or noise artifact and can be dropped 406.
This method can further comprise: determine that 407 whether there is for object detection (for example, to be identified 405 or be lost
Abandon remaining group 406).If it is affirmative, then repetitive operation 404 to 406.If not (it means that all is potential right
As being identified or abandoning), then produce 408 object diagrams.
Example object Figure 50 is shown in Fig. 5.It is right in the presence of four that are numbered as 1 to 4 in object diagram 50 referring to Fig. 5
As.Each object in object 1 to 4 can be by rectangle frame (for example, minimum bounding box or having around minimum bounding box predetermined remaining
The bounding box of amount) it frames.Such rectangle frame can closely cooperate corresponding object, such as in the situation of object 1 to 3, or
Corresponding object is framed with a surplus, such as in the situation of object 4.In addition, a frame can be overlapping with another frame, such as in object 3
In 4 situation, or entirely around another frame, such as in the situation of object 4 and 2.In addition, a frame can be with another frame interval
It opens, such as in the situation of object 1 and 3.These frames can be in order to the subsequent processing to their corresponding objects.Such as explained in Fig. 5
, object 1 can be extracted from object diagram 50 for example, by cutting out its frame along reconstructed hologram image.Then, can pass through
The edge or border detection for executing each embodiment according to the present invention come the edge or boundary 52 of test object 1.Then, edge 52
Readability can be undergone to assess, such as to obtain measurement EC, as described above.
Such as it may be formed at the detailed EC on search range for the measurement EC for enough numbers that multiple depths of focus are collected
Curve 55, as explained in Fig. 5.In EC curve 55, measurement EC can have the bell-shaped profile relative to the depth of focus.Cause
This, can identify good focal plane, for example, optimal or even best focal plane.In some existing methods, searched based on the overall situation
The amplitude of reconstructed image in rope, can also be with forming curves.However, such curve can have apparent peak and valley (for example, such as by
What Figure 20 was explained), and its it is thus provable be difficult to identify optimal focal plane, such as be difficult to determine global maximum.
Fig. 6 shows the intelligibility measure for determining object for explaining each embodiment according to the present invention (for example, object
Readability) method flow chart.It, can be based on the real part of reconstructed image, in Z referring to Fig. 6aWith ZbBetween range in
Depth ZjPlace executes first edge detection 601 for object.Thus, it may be determined that in depth ZjLocate the edge or boundary of object
edgeimage,real, for example, edgeimage,real=conv (real_image, edgeoperator)。
Similarly, this method can further comprise: the imaginary part based on reconstructed image, in depth ZjPlace is executed for object
Second edge detection 602, to obtain edgeimage,imaginary.However, the interchangeable sequence of operation 602 and 601 executes.
This method can further comprise: the first readability of object is obtained based on the standard deviation of first edge detection
603, edgeclearness,real。edgeclearness,realValue can be determined by following formula:
In operation 604, for example, in the mode similar with operation 603, it can be based on the standard deviation of second edge detection
To obtain the second readability edge of objectclearness,imaginary.Similarly, the interchangeable sequence of operation 603 and 604 executes.
In addition, operation 601,603,602,604 can be consecutively carried out, or operation 602,604,601,603 can be consecutively carried out.
Then, the first and second readability value (edge are based onclearness,realAnd edgeclearness,imaginary) measurement
EC for example can determine 605 by following formula:
Wherein
edgeclearness,real=std (| conv (edgeimage,real,gradientoperation) |), and,
edgeclearness,imaginary=std (| conv (edgeimage,imaginary,gradientoperation)|)。
Fig. 7 shows the side for explaining the optimal focal plane for determining one or more objects of each embodiment according to the present invention
The flow chart of method.Method in Fig. 2 is applicable to an object in test, and the method in Fig. 7 is applicable to include one
Or the sample of multiple object of interest.
Referring to Fig. 7, it can receive 701 and reconstruct 201 hologram images.Then, it can recognize and divide in 202 reconstructed images
Object, to obtain object diagram.Then, 704 object of interest can be extracted from object diagram.
In addition, the real and imaginary parts of reconstructed image can be based respectively on, the depth in a range is for extracted
Object executes edge detection 203.Edge detection process may include relative to the first edge detection of real part and relative to imaginary part
Second edge detection.In addition, as previously discussed, can concurrently or sequentially execute first edge detection and second edge detection.
In serial, first edge detection can execute before or after second edge detects.
204 can be assessed based on the standard deviation of each of the first and second edge detections in the depth object
Readability.
Then, it may be determined that whether 707 will be in the readability of another depth assessment object.If it is certainly, then can weigh
203,204 and 707 are operated again.If it is not, then whole to the assessment of the readability of object at different depth in range
It completes.Obtain maximum readability depth can be identified 206 be operation in object optimal focal plane.
It is then determined with the presence or absence of another pair as still to be assessed in 709 reconstructed images.If it is certainly, then repeat
Operation 704,203,204,707,206 and 709.If not (it means that all object of interest have been evaluated), then may be used
Generation 205 includes the image of the information on the optimal focal plane about object.
Fig. 8 shows the optimal focal plane for determining one or more objects for explaining each embodiment according to the present invention
Method another flow chart.All objects being applicable in test as with the method shown in figure 3, and the method in Fig. 8
Be applicable to include one or more object of interest sample.
Referring to Fig. 8,704 objects can be extracted from object diagram.It then, can be based on the real and imaginary parts of reconstructed image, in Za
With ZbBetween the first range in a predetermined depth at for object execute first stage edge detection 303.First stage
Edge detection may include first edge detection associated with real part and second edge associated with imaginary part detection.
In addition, the standard deviation of the real and imaginary parts at the edge of object at a predetermined depth can be based respectively on, to comment
Estimate 304 at a predetermined depth object first stage readability.
Then, it may be determined that 707 objects that whether will be assessed in another depth in reconstructed image.If it is affirmative, then
Repeatable operation 303,304 and 707.If it is not, then can be based on first stage readability associated with predetermined depth
Position, to determine that 805 fall in the first range, Za'With Zb'Between the second range.
Then, 305 second stage edge detection (examples can be executed for object at the different depth in the second range
Such as, the amplitude based on reconstructed image).Then, the second stage that 306 objects can be executed based on second stage edge detection is clear
Clear degree.
It then can determine whether 808 by another depth in the second range determine the object.If it is certainly, then may be used
Repetitive operation 305,306 and 808.If it is not, then the depth for obtaining maximum second stage readability in the second range can quilt
Mark 809 is optimal focal plane.
Then, it may be determined that in 709 reconstructed images with the presence or absence of for assessment another pair as.If it is certainly, then may be used
All operations that repetition is described above in association with Fig. 8.If it is not, then producing 205 includes on the optimal focal plane about object
Information image.
Fig. 9 to Figure 19 is the schematic diagram for illustrating the method for the optimal focal plane of determination of each embodiment according to the present invention.
Different from the use of the method for global search, less depth such as can be used by the method that Fig. 9 to Figure 19 is explained to carry out
Search to optimal focal plane.Therefore, this method accelerates the process for determining optimal focal plane.Used depth number may
It is not enough to be formed EC curve, but disposes these depth by just suitable on search range, the gradient of EC curve can be disclosed, and
And therefore it is contemplated that the narrower range that optimal depth can be fallen into.According to each embodiment, N number of depth can be used to divide search range
At N+1 region, for example, each region has basically the same section, wherein N is the stringent positive number of nature.Such as by Fig. 9
In the embodiment explained to Figure 19, the first range Z can be usedaTo ZbIn three predetermined checkpoints or depth Z1、Z2And Z3。
As discussed above, by obtaining the standard deviation of the real and imaginary parts of target edges respectively and by these standard deviations
Subtractive combination at single scalar value, for example, according toIt can get as joined
The regular curve for describing and explaining according to Figure 21.The systematicness of curve is convenient for diminution search range.
Referring to Fig. 9, checkpoint Z1、Z2And Z3The section of rule can be separated, and range Z can be included inaTo ZbIn.Cause
This, checkpoint Z1、Z2And Z3It can correspond respectively to ZaWith ZbBetween the first quartile of depth, the second quartile is (in or
Value) and third quartile.It is defined as above with checkpoint Z1、Z2And Z3Associated first measurement can be labeled respectively
For ECZ1、ECZ2And ECZ3.It can be assumed that (measuring) the first measurement of side for example, corresponding to maximum first in best focal plane
Strictly increasing, and strictly decreasing is measured the first of the other side, from there through the first measurement EC in the first range of assessmentZ1、
ECZ2And ECZ3Value, it may be determined that the second range that optimal focal plane is likely located at.As a result, region of search can be from biggish
First range shorter is to lesser second range, this is convenient for the search to optimal focal plane.
It is shown to explain, the first measurement ECZ1、ECZ2And ECZ3Among the first metric of maximum shown by black circle
Out, and other first measurement shown by white circle.Referring to Fig. 9, ECZ1It is the first measurement of maximum in this example, and ECZ2
Less than ECZ3.Due to Z1With Z2Between line slope be greater than Z2With Z3Between line slope, therefore by Z1、Z2And Z3It is formed
Curve be intended to decline and it is non-increasing.It gives above-mentioned it is assumed that Z1With Z2Between strictly decreasing region (is shown) with solid arrow can be dark
Show ZaWith Z1Between strictly increasing region (is shown) with dotted arrow.Thus it can determine optimal focal plane (for example, being located at adjacent to most
Good focal plane) fall in ZaWith Z2Between the second range in.
Referring to Fig.1 0, ECZ1It is the first measurement of maximum, and ECZ2Greater than ECZ3.Similarly, Z1With Z2Between strictly decreasing area
Domain implies ZaWith Z1Between strictly increasing region.Accordingly, it can be determined that optimal focal plane falls in ZaWith Z2Between the second range
It is interior.
In addition, referring to Fig.1 1, ECZ1Maximum, and ECZ2Equal to ECZ3.Similarly, it may be determined that optimal focal plane falls in ZaWith Z2
Between the second range in.
From the exemplary analysis of each embodiment explained above based on Fig. 9 into Figure 11, it is noted that the first measurement of maximum is true
Fixed lesser second range or the second range can be associated with maximum first measurement.
Referring to Fig.1 2, ECZ2It is the first measurement of maximum, and ECZ1Less than ECZ3.Strictly increasing (shows) region position with solid arrow
In Z1With Z2Between, and strictly decreasing (is shown) region with another dotted arrow and is located at Z2With Z3Between.Determine that optimal focal plane is fallen in
Z1With Z3Between the second range in.
Referring to Fig.1 3, ECZ2It is the first measurement of maximum, and ECZ1Greater than ECZ3.Similarly, strictly increasing region is located at Z1With Z2
Between, and strictly decreasing region is located at Z2With Z3Between.Determine that optimal focal plane falls in Z1With Z3Between the second range in.
Referring to Fig.1 4, ECZ2It is the first measurement of maximum, and ECZ1Equal to ECZ3.Similarly, strictly increasing region is located at Z1With Z2
Between, and strictly decreasing region is located at Z2With Z3Between.Determine that optimal focal plane falls in Z1With Z3Between the second range in.
Referring to Fig.1 5, ECZ3It is the first measurement of maximum, and ECZ1Less than ECZ2。Z2With Z3Between strictly increasing (with solid arrow
Showing) region can imply that Z3With ZbBetween strictly decreasing region (is shown) with dotted arrow.Determine that optimal focal plane falls in Z2With Zb
Between the second range in.
Referring to Fig.1 6, ECZ3It is the first measurement of maximum, and ECZ1Greater than ECZ2.Due to Z2With Z3Between the slope of line be greater than
Z1With Z2Between line slope, therefore by Z1、Z2And Z3The curve of formation tends to rise up and non-decreasing.Similarly, Z2With Z3
Between strictly increasing region imply Z3With ZbBetween strictly decreasing region.Determine that optimal focal plane falls in Z2With ZbBetween
In second range.
Referring to Fig.1 7, ECZ3It is the first measurement of maximum, and ECZ1Equal to ECZ2.Similarly, Z2With Z3Between strictly increasing area
Domain implies Z3With ZbBetween strictly decreasing region.Determine that optimal focal plane falls in Z2With ZbBetween the second range in.
Referring to Fig.1 8, the EC being equal to each otherZ1And ECZ2Greater than ECZ3。Z2With Z3Between strictly decreasing (being shown with solid arrow)
Region implies that strictly increasing (is shown) region with dotted arrow and (shown) region with another dotted arrow there are also strictly decreasing and is located at
Z1With Z2Between.Determine that optimal focal plane falls in Z1With Z2Between the second range in.
Referring to Fig.1 9, the EC being equal to each otherZ2And ECZ3Greater than ECZ1。Z1With Z2Between strictly increasing (being shown with solid arrow)
Region implies that strictly increasing (is shown) region with dotted arrow and (shown) region with another dotted arrow there are also strictly decreasing and is located at
Z2With Z3Between.Determine that optimal focal plane falls in Z2With Z3Between the second range in.
In second aspect, the present invention relates to the calculating equipment for being used to determine focal plane in a kind of autofocus system, the meters
Calculate the method that equipment is programmed to carry out each embodiment according to the first aspect of the invention.Calculating equipment may include storage
Device, one or more processors and storage in memory and are configured for one be performed by one or more processors
A or multiple programs.
Various embodiments of the present invention may also refer to a kind of autofocus system, which includes surveying for direction
The light source of sample radiation light in examination and the imager of the hologram for acquiring the sample.The autofocus system can be into one
Step includes a kind of calculating equipment for being used to determine focal plane of each embodiment according to the present invention.
Fig. 1 shows the schematic diagram of the autofocus system 100 of each embodiment according to the present invention.Autofocus system
100 can be used for assisting automatic Observation function, such as tracking cell.Specifically, autofocus system 100 can determine for number
Focal plane (for example, focal length parameter) in holographic reconstruction method, so that such reconstructing method can be used to combine identified coke
Plane generates clearly reconstructed image.
Referring to Fig.1, according to 10 He of imaging system that the autofocus system 100 of each embodiment may include according to each embodiment
Calculate equipment 20.In some embodiments, imaging system 10 may include providing the holographic system of one or more holograms of sample
System, and calculate equipment 20 and may include but be not limited to the computer for being configured to handle the hologram from imaging system 10.
Imaging system 10 may include light source 11 and imager 14.Light source 11 can be towards 12 radiant light of sample in test, example
Such as, at least partly coherent light.When light propagation can disclose sample 12 by transmission, scattering and the diffraction characteristic of 12 Shi Youguang of sample
Optical properties.Various embodiments of the present invention can be related to a kind of imaging of transmission hologram for being configured to collecting sample 12 as a result,
System 10.However, various embodiments of the present invention may also refer to a kind of imaging of reflection hologram for being configured to collecting sample 12
System.
Imager 14 can record the optical information (such as wavefront) of the plural form of object in sample 12.Imaging system 10 can
Generate image 15, it may for example comprise the hologram or hologram image of optical information.Image 15 may include can not until being reconstructed
For visual inspection (for example, for by human viewer directly and simple visual explanation) one group of initial data.In addition, at
Equipment 20 is calculated to be used for subsequent processing, as will be discussed further as image 15 can be supplied to by system 10.
In some embodiments, light source 11 may include laser light source or light emitting diode (LED) light source.In addition, sample 12
It may include one or more microbial cells or one or more semiconductor subassembly features.(it runs through for these cells or feature
The disclosure is referred to alternatively as object) it can be focused at different depths or in different focal planes.
Calculating equipment 20 may include processor 21 and memory 22.In some embodiments, processor 21 is central processing
A part of unit (CPU) or computing module.Processor 21 can be configured to execute one or more be stored in memory 22
A program, to execute the specific optimal focal plane operated to determine object in sample 12.Accordingly, in response to from imaging system
The image 15 of system 10 calculates equipment 20 and produces image 16, which includes about one or more objects in sample 12
Optimal depth or optimal focal plane on information.Meter is discussed in detail above in association with each embodiment of the first aspect of the present invention
Calculate the operation or function of equipment 20.
Although using software in calculating equipment 20 in some embodiments, alternatively can be used in other embodiments
Hardware.Hardware realization may be implemented higher performance compared with software realization, but design cost with higher.For real-time
Using generally selecting hardware realization due to rate request.It should be noted that operate as described herein, function and technology can be at least
Partly realized in hardware, software, firmware or any combination thereof.For example, various sides according to various embodiments of the present disclosure
Face may be implemented in one or more processing units, including one or more microprocessing units, digital signal processing unit
(DSP), specific integrated circuit (ASIC), field programmable gate array (FPGA) or any other is equivalent integrated or discrete patrol
Collect circuit system and any combination of this class component.Term " processor ", " processing unit " or " process circuit system " is general
It may refer to individually or combine any foregoing logic circuitry system or any other equivalent circuit of other logic circuitries
System.One or more technologies of the disclosure also can be performed in control unit including hardware.
According in some embodiments of the present disclosure, memory 22 may include any computer-readable medium, including but not
It is programmable only to be limited to random access memory (RAM), read-only memory (ROM), programmable read only memory (PROM), erasable type
Read memory (EPROM), electrically erasable formula programmable read only memory (EEPROM), flash memory, hard disk, solid state drive (SSD), pressure
Contracting disk ROM (CD-ROM), floppy disk, tape, magnetic medium, optical medium or other computer-readable mediums.In certain implementations
In example, memory 22 is included into processor 21.
In some embodiments, calculating equipment 20 may include one or more processors 21, and/or is stored in memory 22
In and be configured for the one or more programs executed by one or more processors 21.One or more programs may include using
In the instruction performed the following operation: reconstructed hologram picture identifies the object in reconstructed image, the real part based on reconstructed image
First edge detection is executed for object in a depth, the imaginary part based on reconstructed pattern is executed in the depth for object
Second edge detection assesses the first readability and/or base in the depth object based on the result of first edge detection
The second readability in the depth object is assessed in the result of second edge detection.
In addition, one or more programs may include for performing the following operation in each embodiment according to the present invention
Instruction: reconstructed hologram picture identifies the object in reconstructed image, and the real part based on reconstructed image is pre- in the first range
First stage edge detection is executed for object at depthkeeping degree set, the imaginary part based on reconstructed image is pre- in the first range
First stage edge detection is executed for object at depthkeeping degree set, is based on first stage associated with real and imaginary parts edge
The result of each of detection assesses the first stage readability of the object at predetermined depth set, and/or based on the
One stage readability identifies the second range associated with optimal focal plane in the first range.
In the third aspect, the invention further relates to a kind of computer program product, which is used for when in root
According to each reality executed when being executed in the calculating equipment of each embodiment of the second aspect of the present invention according to the first aspect of the invention
The method for applying example.
Claims (15)
1. a kind of for determining the auto focusing method of the focal plane of at least one object, which comprises
The hologram image of at least one object is reconstructed in order to provide the reconstructed image at multiple and different depths of focus,
Wherein the reconstructed image includes real component and the imaginary component for jointly encoding phase and amplitude information；
First edge detection is executed to the real component at least two depth in the multiple different focal point depth, and
And it is detected at least two depth to execute second edge to the imaginary component；
It is obtained based on the first statistic dispersion measurement detected relative to the first edge every at least two depth
First intelligibility measure of one, and based on the second statistic dispersion measurement detected relative to the second edge to obtain
State second intelligibility measure of each of at least two depth；And
The coke of at least one object is determined based on the comparison of the clarity scalar measurement of at least two depth
Plane,
Wherein the scalar measurement is based on first intelligibility measure and second intelligibility measure.
2. the method as described in claim 1, which is characterized in that further comprise in the mark reconstructed image it is described extremely
A few object,
Wherein following steps are applied to correspond to the regional area of at least one object in the reconstructed image:
Execute the first edge detection and second edge detection；
Obtain first intelligibility measure and second intelligibility measure；And
Determine the focal plane of at least one object.
3. method according to claim 2, which is characterized in that at least one object packet in the mark reconstructed image
It includes:
Digitize the reconstructed image；
Identify the connected region with same numbers value；And
Divide each of described connected region.
4. method according to claim 2, which is characterized in that
Multiple objects are identified,
Multiple focal planes corresponding to the multiple object are determined, and
Wherein, the method further includes: the image-region of each of the multiple object will be corresponded to corresponding burnt
It is stitched together in plane to form composograph, the composograph includes each of the multiple object.
5. the method as described in claim 1, which is characterized in that the first statistic dispersion measurement or second statistic dispersion
Measurement includes standard deviation.
6. method as claimed in claim 4, which is characterized in that obtain first intelligibility measure or obtain described second clearly
Clear degree measures
Obtain the gradient magnitude of the result of the first edge detection or second edge detection；And
Obtain the standard deviation value of the gradient magnitude.
7. the method as described in claim 1, which is characterized in that described at least two in the multiple different focal point depth are deep
Degree includes each depth being evenly distributed in preset range.
8. the method for claim 7, which is characterized in that each deep packet being evenly distributed in the preset range
Include first quartile, the second quartile and the third quartile of the preset range.
9. the method as described in claim 1, which is characterized in that further comprise:
The further depth of at least one of the multiple different focal point depth is determined based on identified focal plane；
The first edge detection and second edge detection are executed at least one described further depth；
Obtain the first intelligibility measure and the second intelligibility measure of at least one further depth；And
The focal plane is adjusted based on the clarity scalar measurement of at least one further depth.
10. the method as described in claim 1, which is characterized in that further comprise:
The further depth of at least one of the multiple different focal point depth is determined based on identified focal plane；
It is directed at least one described further depth based on the amplitude of the reconstructed image and executes second stage edge detection；
And
The second stage intelligibility measure of at least one object is assessed based on the result of the second stage edge detection.
11. method as claimed in claim 9, which is characterized in that
At least two uniform depth is distributed in the first predetermined depth range,
At least one described further depth includes each depth being evenly distributed in the second depth bounds,
Second depth bounds are narrower than the first predetermined depth range, and
Second depth bounds are the comparisons of the clarity scalar measurement based at least two depth come really
Fixed.
12. the method as described in claim 1, which is characterized in that
Execute first edge detection and the second edge detect include: by the real component or the imaginary component respectively with
Laplce's mask carries out convolution.
13. a kind of calculating equipment for the focal plane in autofocus system for determining at least one object, the calculating
Equipment is programmed to carry out a kind of method, which comprises
The hologram image of at least one object is reconstructed in order to provide the reconstructed image at multiple and different depths of focus,
Wherein the reconstructed image includes real component and the imaginary component for jointly encoding phase and amplitude information；
First edge detection is executed to the real component at least two depth in the multiple different focal point depth, and
And it is detected at least two depth to execute second edge to the imaginary component；
It is obtained based on the first statistic dispersion measurement detected relative to the first edge every at least two depth
First intelligibility measure of one, and based on the second statistic dispersion measurement detected relative to the second edge to obtain
State second intelligibility measure of each of at least two depth；And
The coke of at least one object is determined based on the comparison of the clarity scalar measurement of at least two depth
Plane,
Wherein the scalar measurement is based on first intelligibility measure and second intelligibility measure.
14. calculating equipment as claimed in claim 13, which is characterized in that the autofocus system includes:
It is configured to the light source towards sample and image radiation light in test；And
It is configured to acquire the imager of the hologram of the sample.
15. a kind of non-transitory computer-readable medium for being stored thereon with instruction, described instruction can be executed by processor to execute
Method for determining the focal plane of at least one object, which comprises
The hologram image of at least one object is reconstructed in order to provide the reconstructed image at multiple and different depths of focus,
Described in reconstructed image include real component and imaginary component for jointly encoding phase and amplitude information；
First edge detection is executed to the real component at least two depth in the multiple different focal point depth, and
And it is detected at least two depth to execute second edge to the imaginary component；
It is obtained based on the first statistic dispersion measurement detected relative to the first edge every at least two depth
First intelligibility measure of one, and based on the second statistic dispersion measurement detected relative to the second edge to obtain
State second intelligibility measure of each of at least two depth；And
The coke of at least one object is determined based on the comparison of the clarity scalar measurement of at least two depth
Plane, wherein the scalar measurement is based on first intelligibility measure and second intelligibility measure.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW103141335A TWI557462B (en) | 2014-11-28 | 2014-11-28 | Autofocus system and method |
| PCT/EP2015/078089 WO2016083620A1 (en) | 2014-11-28 | 2015-11-30 | Autofocus system and method in digital holography |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108377658A CN108377658A (en) | 2018-08-07 |
| CN108377658B true CN108377658B (en) | 2019-10-18 |
Family
ID=54771100
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201580065121.7A Active CN108377658B (en) | 2014-11-28 | 2015-11-30 | Autofocus system and method in digital holography |
Country Status (7)
| Country | Link |
|---|---|
| US (1) | US10241471B2 (en) |
| EP (1) | EP3224678B1 (en) |
| JP (1) | JP6490219B2 (en) |
| CN (1) | CN108377658B (en) |
| CA (1) | CA2968024C (en) |
| TW (1) | TWI557462B (en) |
| WO (1) | WO2016083620A1 (en) |
Families Citing this family (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3260841B1 (en) * | 2016-06-22 | 2024-03-13 | Uponor Oyj | Detecting microscopic objects in fluids |
| US11373278B2 (en) * | 2016-09-30 | 2022-06-28 | University Of Utah Research Foundation | Lensless imaging device |
| EP3339963B1 (en) * | 2016-12-21 | 2020-08-12 | IMEC vzw | An apparatus and a method for in-line holographic imaging |
| JP6969164B2 (en) * | 2017-05-31 | 2021-11-24 | 株式会社リコー | Evaluation device, evaluation program and evaluation method |
| WO2019102272A1 (en) * | 2017-11-24 | 2019-05-31 | Sigtuple Technologies Private Limited | Method and system for reconstructing a field of view |
| CN108364296B (en) * | 2018-02-09 | 2020-12-01 | 重庆东渝中能实业有限公司 | Cell population space distribution construction method based on multilayer holographic reconstruction and focusing strategy |
| US10699875B2 (en) * | 2018-11-13 | 2020-06-30 | Fei Company | Confocal imaging technique in a charged particle microscope |
| US12130588B2 (en) * | 2019-10-11 | 2024-10-29 | miDiagnostics NV | System and method for object detection in holographic lens-free imaging by convolutional dictionary learning and encoding with phase recovery |
| EP3839636B1 (en) * | 2019-12-20 | 2024-03-13 | Imec VZW | A device for detecting particles in air |
| CN112098997B (en) * | 2020-09-18 | 2021-10-15 | 欧必翼太赫兹科技(北京)有限公司 | Three-dimensional holographic imaging security inspection radar image foreign matter detection method |
| CN112529791B (en) * | 2020-11-16 | 2023-05-26 | 中国海洋大学 | Adaptive multi-focus restoration method based on plankton digital holographic image |
| CN112969026A (en) * | 2021-03-18 | 2021-06-15 | 德州尧鼎光电科技有限公司 | Focal plane automatic focusing method of imaging ellipsometer |
| CN113286079B (en) * | 2021-05-10 | 2023-04-28 | 迈克医疗电子有限公司 | Image focusing method and device, electronic equipment and readable storage medium |
| CN113359403B (en) * | 2021-05-21 | 2022-08-12 | 大连海事大学 | Automatic focusing method for lens-free digital holographic imaging |
| CN114548884B (en) * | 2022-04-27 | 2022-07-12 | 中国科学院微电子研究所 | Package identification method and system based on pruning lightweight model |
| CN116642895B (en) * | 2023-07-27 | 2023-09-29 | 宏椿智能科技(苏州)有限公司 | Five-face imaging 3D reconstruction method for focal plane sweep |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101467087A (en) * | 2006-06-09 | 2009-06-24 | 维谷设备有限公司 | Method and apparatus for auto-focussing infinity corrected microscopes |
| CN103383513A (en) * | 2012-05-04 | 2013-11-06 | 三星电子株式会社 | Focus detecting apparatus |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2011160068A1 (en) * | 2010-06-17 | 2011-12-22 | Purdue Research Foundation | Digital holographic method of measuring cellular activity and measuring apparatus with improved stability |
| WO2012099220A1 (en) * | 2011-01-21 | 2012-07-26 | 兵庫県 | Three-dimensional shape measurement method and three-dimensional shape measurement device |
| CN105264443B (en) * | 2013-06-06 | 2019-05-31 | 视瑞尔技术公司 | The device and method of data for computed hologram |
| KR101820563B1 (en) * | 2014-12-31 | 2018-01-19 | 한국전자통신연구원 | Data format for hologram, and apparatus and method for holographic video system |
-
2014
- 2014-11-28 TW TW103141335A patent/TWI557462B/en not_active IP Right Cessation
-
2015
- 2015-11-30 US US15/528,202 patent/US10241471B2/en active Active
- 2015-11-30 CN CN201580065121.7A patent/CN108377658B/en active Active
- 2015-11-30 CA CA2968024A patent/CA2968024C/en not_active Expired - Fee Related
- 2015-11-30 EP EP15804106.1A patent/EP3224678B1/en active Active
- 2015-11-30 JP JP2017528223A patent/JP6490219B2/en not_active Expired - Fee Related
- 2015-11-30 WO PCT/EP2015/078089 patent/WO2016083620A1/en active Application Filing
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101467087A (en) * | 2006-06-09 | 2009-06-24 | 维谷设备有限公司 | Method and apparatus for auto-focussing infinity corrected microscopes |
| CN103383513A (en) * | 2012-05-04 | 2013-11-06 | 三星电子株式会社 | Focus detecting apparatus |
Also Published As
| Publication number | Publication date |
|---|---|
| CA2968024C (en) | 2019-11-12 |
| CA2968024A1 (en) | 2016-06-02 |
| JP6490219B2 (en) | 2019-03-27 |
| TW201619652A (en) | 2016-06-01 |
| US10241471B2 (en) | 2019-03-26 |
| TWI557462B (en) | 2016-11-11 |
| EP3224678A1 (en) | 2017-10-04 |
| WO2016083620A1 (en) | 2016-06-02 |
| US20170322516A1 (en) | 2017-11-09 |
| JP2017538160A (en) | 2017-12-21 |
| EP3224678B1 (en) | 2019-02-27 |
| CN108377658A (en) | 2018-08-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108377658B (en) | Autofocus system and method in digital holography | |
| US11721018B2 (en) | System and method for calculating focus variation for a digital microscope | |
| CN110274877B (en) | 3D spectral imaging system and method based on scattering medium | |
| JP5001286B2 (en) | Object reconstruction method and system | |
| JP6752212B2 (en) | Image system and how to use it | |
| CN113960908B (en) | Holographic method for characterizing particles in a sample | |
| CN107003229B (en) | Analytical method comprising holographic determination of the position of a biological particle and corresponding device | |
| CN110214290A (en) | Microspectrum measurement method and system | |
| US20130147919A1 (en) | Multi-View Difraction Grating Imaging With Two-Dimensional Displacement Measurement For Three-Dimensional Deformation Or Profile Output | |
| US20220351347A1 (en) | Computational refocusing-assisted deep learning | |
| CN110992363A (en) | Three-dimensional magnetic flux leakage detection defect reconstruction method based on generation countermeasure network | |
| JP6058902B2 (en) | Skin condition analysis method, skin condition analysis apparatus, skin condition analysis system, program for executing the skin condition analysis method, and recording medium recording the program | |
| EP3992609A1 (en) | Image processing device, evaluation system, image processing program, and image processing method | |
| Varjo et al. | All-in-focus image reconstruction from in-line holograms of snowflakes | |
| Volynskiĭ et al. | Recursion algorithms for processing video information in optical-coherence-tomography systems | |
| US20230274441A1 (en) | Analysis method and analysis apparatus | |
| Mendes et al. | Three-dimensional surface reconstruction for evaluation of the abrasion effects on textile fabrics | |
| JP5993587B2 (en) | Skin condition analysis method, skin condition analysis apparatus, skin condition analysis system, program for executing the skin condition analysis method, and recording medium recording the program | |
| Chen et al. | Shape reconstruction from focus and image fusion | |
| Harding | Focus-Based Optical Metrology | |
| El Mallahi et al. | Automated 3D detection of particles using digital holographic microscopy with partially coherent source-Applications in environmental monitoring | |
| Lindblad et al. | Optimizing optics and imaging for pattern recognition based screening tasks | |
| Sierra et al. | Reconstructing features of thick objects from phase images | |
| Caldeira | Development of bioinformatics tools to track cancer cell invasion using 3D in vitro invasion assays |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20211231 Address after: Leuven Patentee after: IMEC VZW Address before: Taiwan, Hsinchu, China, Hsinchu Science Industrial Park, 6, 2, 4 Patentee before: IMEC TAIWAN Co. Patentee before: IMEC VZW |
|
| TR01 | Transfer of patent right |